Hacker News new | ask | show | jobs
by nzoschke 16 days ago
> None of the music videos were great

Glad they acknowledge this.

Curious how much time in addition to tokens this costs. If you have to spend $25 and wait 45 minutes to get a basically unwatchable video, I'm not worried about indie film makers being replaced just yet...

4 comments

"not great" is a huge understatement
It's sickening we're running out of compute for this shit lol. I don't blame the author though, it is what it is.
If it wasn't this it'd be for mining shitcoins.
This wasn’t possible even a year ago, with the speed of things changing and how much money is spent on movies is there really a doubt that someone will be able to make $100 million movie for less than $1 million in token spend?
> This wasn’t possible even a year ago

Just because it’s possible now, doesn’t mean it’s worth doing.

Do you... Have a lot of professional experience with making $100 million-budget movies?
Even small budget indie flicks, when filmed in LA, have a ridiculous amount of paperwork and dealing with unions. Which isn't surprising, since it's an established industry there. Spending $1 million for each day of filming is totally achievable. I don't have personal experience with this but I know some people in the industry and have had conversations with them. Was background/extra in a movie. An AI director/producer doesn't have to deal with the human aspect, actor and other people's egos and clashing behavior. There are moments of film gold that are the result of human actor's human personalities, so we'd lose out on that though.
I'm classically trained and I honestly can't really tell how these are worse than the human produced ones. They all look kinda the same to me.
Classically trained in music videos??
Presumably they went to college for it, yeah. Film schools, eg https://cinema.usc.edu/ or https://tisch.nyu.edu/
No, as in classical music. The analog kind.
So, how is that relevant to the part that is generated by AI?
I honestly can't tell a good music video from a bad one. I don't know what makes a music video "good". These AI generated ones seemed pretty good to me, of the same caliber as those Michael Jackson or "Beez in the trap" things. It's all just butt shaking and sunglasses in Philips Hue lights chanting some meaningless words.
Imagine the blogpost was on AI generated classical music. Everyone, including the authors, agrees the result is shit. You, being classically trained, can understand more than most why the results are bad.

Then someone comes along and says “I honestly can't really tell how these are worse than the human produced ones. They all sound kinda the same to me” and “I honestly can’t tell a good piece of music from a bad one. I don’t know what makes a piece of music “good”. These AI generated ones seemed pretty good to me, on the same caliber as those Mozart or Beethoven things”.

That’s what you’re doing, and that’s why you’re being downvoted. It’s fine that you can’t tell, but it’s supremely obvious to everyone else.

As opposed to the non-analog kind??
I'm seeing these type of comments a lot more. A type of comment where it seems like the "user" (commentor) makes a categorically wrong comment about the linked article that would be sort of right if it were a different medium. Clearly whatever was used here triggered on the "music" part, but didn't understand what the article was really about. Typically it goes beyond just not reading the article (as in this case)

I'd assume that it was a bad LLM take, but it looks like the rest of the comments are normal.

“Classically trained” in the musical context usually means “I took a year of piano lessons.”
It's just categorically a wrong statement. Just because someone knows how to play a musical instrument, I'm not sure why it would make any sense that they would understand what makes a good music video - they are a completely different medium.

You might be able to discuss asthetics from a music theory standpoint because you are "classically trained", again, it is just a weird mismatch that stood out for me recently, but I think everyone is a little more paranoid lately.

Its like misunderstanding the concept of "music video" as one form music can take, presumably alongside "music audio," and the being "classically trained" is meant to speak to a strength in "music" in general.
Your dislike of music videos in general has meant that you haven’t had the interest to discern the good from the bad. There are plenty of high quality music videos. It’s like someone saying “I’m not really into classical music, it sounds all the same to me”. If they had an interest they’d know that it doesn’t sound all the same. So what’s the point in saying so?
Directors and editors using Seedance can fire the film studio.

This is a fundamental shift in how storytelling is funded and made, not in who does the driving.

Same as is happening with code.

>funded and made

I’m curious (admittedly skeptical) what you mean by this. Are you talking about a world where director’s just like…don’t actually make movies and create AI media?

Same thing.
[flagged]
Dude you have been salivating at the thought of seeing developers fired, now film studios, for months now. Give it a rest, this is unhealthy.
> Dude you have been salivating at the thought of seeing developers fired,

I don't think you understand my arguments at all.

I don't want people to lose jobs. I don't want the advantages to accrue to a few hyperscalers.

I want there to be more work, more money, more societal progress. For everyone.

AI currently enables experts in their domains (code, film, game design, music) to get the work of small teams done. It is now possible for dedicated, hard-working ICs in several domains to produce extremely good products in a short amount of time without external capital and without studio meddling. This is a good thing.

Like your other comment this is so general I’m not entirely sure what you’re describing.

AI in production has allowed us to salvage bad audio and fix mistakes/not go back for re-shoots or ADR when it’s acceptable. It’s currently a repair/salvage and corner cutting tool. As for video, generative AI is not creating particularly usable stuff and it’s at best useful for short bursts.

If you’re in the commercial game it’s a little more useful - a few seconds of a generic looking kitchen sink and faucet running isn’t a big ask. But also getting quality B-roll isn’t particularly laborious as it is.

It’s great for transcriptions/captioning first passes but we already had great tools for that prior to the ChatGPT era.

It’s not cutting for us, it’s not generating useful shots most of the time and only for a few seconds, it’s not cutting out middlemen. I don’t see what you’re seeing.

> I don't want people to lose jobs. I don't want the advantages to accrue to a few hyperscalers.

Versus:

> It is now possible for dedicated, hard-working ICs in several domains to produce extremely good products in a short amount of time without external capital and without studio meddling

A contradiction exists: lower-capital requirements means less employment all-around and more profit to fewer people.